Re: Generally accepted cut-off limit for -doc packages
[ Not particularly replying to Joey, just using his mail as a jump-in point ]
Martin Schulze <email@example.com> writes:
> Andreas Tille wrote:
>> Sorry for my ignorance but I thought that there is a reference for his
>> rejection statement. I just made sure that the maintainer which I volunteered
>> to sponsor builded the package compliant to debian-policy 22.214.171.124 which says:
>> If a package comes with large amounts of documentation which many
>> users of the package will not require you should create a separate
>> binary package to contain it, so that it does not take up disk space
>> on the machines of users who do not need or want it installed.
> It seems to me that this limit is not well documented. However, since
> James in his incarnation of the ftpmaster has the final say, I'd
> suppose to adjust both the developers reference and the
> new-maintainers guide to mention what Oliver Elphick said
I don't know where people get the idea that I was dictating anything;
the .reason file (which is 0644 on auric and was already quoted in
this thread) said: "Please consider merging the loki-doc package back
into loki?" Notice how it's a question? I wasn't trying to have the
final say or just about anything else that people have been suggesting
in this thread.
I also didn't just pull the 500k limit at random out of my evil cabal
hat, it's what I honestly remember as being consensus from a
discussion many years ago (IIRC, initially WRT cvs(-doc) back when Tom
Lees first took it over). I couldn't find it documented but I a)
didn't try very hard, b) knew a lot of this kind of documentation got
lost with the packaging manual and thought this might just be another
example of that.
My reply to Andreas was "tense" because if it hadn't been for the fact
that he replied to a role address, I wouldn't have seen his mail as
he's currently in my twitfile.
Off the top of my head and in the recent past I've seen:
o A source package with multiple (> 6) binaries and each one was
packaged as a separate deb for no particular reason.
o A 7k xmms plugin in a separate binary package (with no other files
modulo the policy mandated ones).
o 8k/22k (deb/installed) -examples package split out of a large C++
o A separate -demo package with a single 2k file (modulo policy mandated).
o A separate -doc package with 17 files and a grand total 70k of docs.
So yes, I do tend to REJECT what are IMO less-than-ideal package
splits if only to encourage the maintainer to think about the pro's
and con's of package splitting. I try to carefully phrase them (and
differentiate them from e.g. "you have no license in the copyright
file, fix it, you tramp." type REJECTs) to indicate they're not a "no,
not ever" REJECT but a "fix this please and/or explain what I'm
missing" REJECT. I guess maybe I'll have to add some sort of template
footer to distinguish between the two, but like the MOTD on auric it's
depressing that it's necessary :(